Shadow

Tags:

Much of this won’t make sense to anybody else besides me. But hopefully it will.

In my view the research on ToM has focused too much on verbal reasoning, which has its root in the Lewis-type philosophy that deals only with truth values. I believe action speaks louder than words, under circumstances, of course.

From a simulation theory persepctive, the difficulty with ToM, false belief, and counterfactual is discounting a portion of the valid information while maintaining particular "false" info in reasoning. The above design is based on the assumption that language can help keeping things straightened in this confusing task.

Note that the language in this case is simple decleartive sentences, not the complex, tensed complement Jill DeVillis talked about. When the child is in the study to re-enact, she is INSIDE the simulation. There is no need to — and I argue children won’t — resort to such structure. In other words, the child does not have to be aware that this is a simulation, but only need to be reminded of changes in the rules/conditionals.

This is very different from the traditional ToM tasks, where the child has to be consciously aware — and have explicitly retrieve info from — the simulation and the reality simultanuously.

An essential claim I would like to make is that a child’s mind resides in one world at a time. Anybody who has played with children know how quickly and easily they move in and out othe worlds during pretence. They must have mechanisms readily available to resolve this kind of conflicts between real- and imagined-worlds.

Here is my theory of why young children are natural-born pretenders — not that they are better ones. The idea can be summarized as "less is more."

In a simulation framework. let’s say ones’ behavior is governed by a set of constraints, Ci, such as "dogs bark; they don’t miao." You can think of Ci as constraints in the OT sense or as old-fasioned rules. It doesn’t matter at this point. The agent’s behaviors (or thoughts) are spontaneous but are limited by the Ci. With pretence, we add one or more constraints to the set, C’. The counterfactual constraint set Ci may be expanded as secondary constraints are derived from the existing C’. For example, if I am a dog, then I should at least know that I bark and walk on all fours. Whether or not I realize that dogs never sweat is limited by what I know and whether it’s relevant.

Unless anything in the counterfactual set C’ is in conflict with Ci, the real-world constraints, the agent can wonder about in the old fashioned way. But when C’ colides wich Ci, then interesting decisions have to be made. If you are strictly simulating, C’ should always dominate Ci. For example, if you decided to be a dog, and you walk around on twos, then the right thing to do would be to stop your old habbit and drop down on all fours.

It should be obvious at this point why you and me cannot pretend so well (any more). We know too much. Our knowledge and logic gets us in constant clashes between the known and imagined. If you think a pretended dog has to crawl all the time, you miss the point.

Children, on the other hand, are not worried about all the inhibitors. Having fewer rules to live by, they roam the imagined world freely, only to stop occassionally to resolve conflicts. Alternatively, children may be more willing to live in a world with conflicts and ambiguities.

So what does this mean to the Theory of Mind and counderfactual reasoning? 

• ToM — the theory that an agent has to have a theory that other agents have minds in order to interprete or predict their behaviors –  is not a general framework for understanding children’s pretence, imagination, and behaviors

• Research should shift attention from the verbal/truth-value tradition, and look at cognitive in situ and in motion.  For an alternative idea, see this quick mock up.

• False belief tasks require children to access the imagined world and reality at the same time. This something hard to do, and is NOT ESSENTIAL to undertanding others’ mind or behavior. Take the Sally-Ann task. In the adult interpretation, you’d have to imagine 2 worlds (note that both are imagined — in reality, a doll named Sally cannot hide things!) — one in which you observed the events, and one for Sally, who experience the "same" events but from a different perspective. The Smarty task is similar — the current world and the "old" world.

  The simulation theory would predict the difficulties in switching world views: young children are immersed in one world at a time, and would not build barriers to stop information sharing between worlds. Again, this may be a limitation in cognitive abilities — predicted by Piaget — but it’s in no way limited to mind-thinking.

• The Sally-Ann type task assum strong constraints on behavior, which children may not readily access. For example, we would automatically infer from the story that Sally left the room while Ann moved the candie that Sally didn’t know the new location of the candie, and from there, infer that she would believe the candies were still under the old place.

  I am not sure young children could or would do that. Based on what I know from my 2 children, I am perfectly happy to predict that a child might think:

  "ok, Sally walked out… now she comes back … now she gets the candy."

  This is a pretty good simulation of what happened, except that a few potential conflicts are not resolved or event raised. One of which being that (a) the candies have been moved to a new place, which everybody — including Sally — knows, and (b) Sally still assumes the candies are at the old place, a proposation derived above, which is several steps away from that the child observed.

• The "Alternative Worlds" model cannot be a valid psychological theory, not only because of because the parallelism, but also for some philosophic/semantic reasons. See my post Troubles in the Alternative Worlds .